LM3: Learning Multiscale Material Models
Department of Mechanical Engineering
University of Michigan, Ann Arbor

Thomas D Swinburne, PhD
Assistant Professor, University of Michigan
Visiting researcher, CNRS Physique
Associate Editor, Computational Materials Science
CV  /  Google Scholar  /  GitHub
tswin-at-umich-dot-edu

We push the frontier of multi-scale modelling for metals, connecting atomic dynamics, microstructure and mechanical properties.
Keywords: Dislocations, Irradiation Damage, Phase Diagrams, MD/ML/UQ.

Metallic alloys uniquely combine strength, ductility and reusability due to a complex hierarchy of plasticity mechanisms which are only partially understood. In many settings, prediction of component failure is a largely open problem, with important implications across industry and society.

Fatigue alone is estimated to cost 2-3% of US GDP, while irradiation-induced embrittlement is a key issue for all fission and fusion reactors.

Experiments probing the microstructural origins of failure are typically indirect, destructive and expensive, meaning data is sparse. Predictive models which help diagnose and mitigate failure are thus urgently sought.

Our research blends theory and data to build multiscale models of metal plasticity, often with applications in the nuclear sector. We draw broadly from concepts in applied math, machine learning, physics and mechanics.

We are hiring in Ann Arbor & Paris!
See open positions or email tswin-at-umich-dot-edu. Past members have gone on to obtain competitive positions in academia and industry.

Current research topics
Projects balance theory/applications depending on your interest.

• Application focus:
  • Slip transfer in complex dislocation networks
  • High-dpa irradiation damage for nuclear fusion
  • Exploring phase diagrams and kinetics
  • Mechanistic inverse design of non-dilute alloys
• Theory/method development focus:
  • Model-form uncertainty quantification (UQ) for ML surrogates
  • Data-driven forecasting of long-timescale simulation behaviour
  • End-to-end differentiable simulations for UQ and design
  • Efficient fine-tuning of universal potentials (e.g. MACE)

Selected recent work
See Google Scholar for an up-to-date list.

• Entropy of dislocation glide (Nat. Comm. 2025 / PAFI / Rodney Group)
  Why ML potentials are required to accurately model bcc plastic flow
• Score matched free energies (Nat. Comm. 2025 & AI4Mat / code)
  MLIP phase diagrams as functions, for UQ and inverse problems.
• Implicit differentiation in MD (NPJ 2025 featured / code)
  MLIP energy landscapes as functions, for UQ and inverse problems.
• Misspecification uncertainty in MD (NPJ 2025 / code)
  Using POPS to explore model-form UQ of SNAP and MACE.
  
• POPS model-form UQ for deterministic models (ML:S&T 2025 / code)
  Solving Bayes’ ignorance of model-form UQ (misspecification).
  —-
• MACE foundation model for ML potentials (Csyani Group / arXiv 2024)
  Section A.13. Dislocations are a tough extrapolation test for UMLIPs.
  —-
• Coarse-graining & forecasting atomic simulations ( PRL 2023)
  ML feature vectors used for MLIPs can also forecast complex futures
• Embedding ab initio in MLIPs for solute studies (Acta Mat. 2023 )
  QM/ML for dislocation-solute interactions. Applied to W alloys
  
• A15 defects in fcc (Nat. Comm. 2023 / TAMMBER / Marinica Group)
  Irradiation defects can grow as three-dimensional Laves phase clusters

Recent or upcoming travel and talks
EL2026, Telluride, USA, 07/26
SIAM UQ, Minneapolis, USA, 03/26
Frontiers of Theory and AI in Sustainable Mat. Sci., Dusseldorf, 01/26
USACM UQ, Webinar, USA, 01/26
NeurIPS AI4Mat, San Diego, USA, 12/25
Dislocations 2025, Miami, USA, 11/25
CECAM: UQ from DFT to ML, EPFL, Lausanne, 11/25
CoMPASs workshop, ICMS, Edinburgh, 11/25 (unavailable)
IPAM Electrochemistry, UCLA, 10/25
USACM Nanomechanics 2025, Urbana-Champaign, USA, 09/25